INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED11
15 - 18 AUGUST 2011, TECHNICAL UNIVERSITY OF DENMARK
THE IMPACT OF EXAMPLES ON CREATIVE DESIGN:
EXPLAINING
FIXATION
AND
STIMULATION
EFFECTS
Marine Agogué1, Akin Kazakçi1, Benoit Weil1, Mathieu Cassotti1
1Centre de Gestion Scientifique, Mines ParisTech
ABSTRACT
Converging evidences have indicated that the ability to generate creative ideas could be limited by
recently activated knowledge such as examples of solutions. However, neuroimaging studies have
recently demonstrated that exposure to examples did not systematically lead to fixation and could on
the contrary have a stimulation effect on creativity. Our hypothesis is that there are two types of
examples that C-K theory helps to characterize: (1) restrictive examples that do not change the
definition or the attributes of the object, and (2) expansive examples that modify its identity by adding
unexpected attributes. In two studies, we explored the impact of restrictive and expansive examples on
a creative task. We then hypothesized that the introduction of an example during the task would
provoke participants to propose solutions of a higher originality when the provided example was
expansive. In contrast, when the provided example was restrictive, we hypothesized that the
originality of the solutions would be reduced. Results confirmed that solutions proposed by the group
exposed to restrictive example are less original than those given by groups exposed to expansive
examples.
Keywords: Concept-Knowledge Design Theory, fixation effects, example in creativity
1 INTRODUCTION
Today there are large expectations regarding innovation and creativity, and some sectors seem to be
less innovative than others. Trends in cognitive psychology have clarified some obstacles that most
people are likely to face in creative situations. More specifically, numerous reports have indicated that
recently activated knowledge can constrain the ability to generate creative ideas [1]. For example,
creative problem solving can be inefficient when the solution requires subjects to generate an atypical
object function and when the object’s typical function has been primed [2]. In the psychology
literature this effect is labeled as the functional fixedness or fixation. In the same vein, converging
evidence reports a fixation bias when subjects are asked to generate new ideas after being exposed to
solution examples.
Nevertheless, in the specific field of neuroimaging, recent studies have made significant contributions
to our understanding of the neuro-cognitive influence of ideas produced by other people by
demonstrating an opposing trend [3]. Indeed, in contrast with the fixation phenomena, this work
highlights that creative cognition can be improved effectively by means of the ideas of others.
Although the discrepancies in the results of these studies underscore the need for further and focused
research in this field, little is known about the potential impact of the nature of examples in the realm
creativity. This lack of knowledge is regrettable if one considers that examining the precise influence
of different types of examples on idea generation may provide experimental arguments enabling to
reconcile studies showing either provocative or constraining effects of examples.
In the field of design science, the development of C-K theory modeling creative reasoning [4], [5]
offers a new and rich interpretation of both the fixation and the stimulation effects. Our hypothesis is
therefore that there are two types of examples that C-K theory helps to characterize: (1) restrictive
examples that do not change the definition or the attributes of the object, and (2) expansive examples
that modify the identity of the initial task by adding unexpected attributes. We build therefore a
ICED 11
1
INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED11
15 - 18 AUGUST 2011, TECHNICAL UNIVERSITY OF DENMARK
framework that allows us to characterize the nature of examples that provoke either fixation or
stimulation effects.
The purpose of this paper is to examine the impact on creativity in relation to the nature of the
examples. We do this via a creative task where the aim is to design a way to drop a hen's egg from a
height of 10 m so that it does not break. We choose this familiar task because it requires minimal
engineering expertise and allows many possible solutions. We then use C-K theory to chart responses
that do not include only fixation effects and to characterize the nature of the type of responses that can
be possible solutions. In this way, we will test how restrictive examples and expansive examples
impact the originality of the solutions to the design task.
2
LITERATURE REVIEW
2.1 Fixation and stimulation effect
Over the last six decades, numerous reports in the psychological literature have indicated that the
ability to generate creative ideas could be limited by recently activated knowledge. One of the most
well known examples of this failure in the domain of creative problem solving is the phenomenon of
functional fixedness. Here, people fail to see new ways of using objects that could lead to an original
solution to a problem, because they remain bound to previously activated prescription of the object.
For example, in the ‘candle problem’ [6], [7], subjects are provided with a candle, a book of matches
and a box of tacks, and asked to fix the candle to a vertical screen in such a way that it will burn
without dripping wax on the table below. The solution to this problem requires subjects to use the tack
box in an unfamiliar way: as a platform. Adults failed to spontaneously find this solution when the box
was presented full of tacks. In other words, priming the box’s typical function blocked the subjects’
ability to use the box in an atypical manner. Traditional interpretations of functional fixedness suggest
that accumulated knowledge about the typical properties of the object is automatically activated and
blocks alternative uses of the object.
Another famous example of the fixation effect was reported by Smith, Ward and Schumacher [8], who
observed and described the constraining effects of the examples used in generating creative ideas. In
the classical task, participants were required to imagine and draw animals that lived on another planet
that were very different to those on Earth. Prior to the drawing of the animal, participants are exposed
to examples that have fundamental elements in common (such as eyes, antennas,…). Two independent
assessors noted then whether the subjects’ drawings contained any of these elements. Results showed
that participants tended to incorporate these elements in their own drawing, despite an explicit warning
to avoid replicating features given in the examples. (see also [9]).
Interestingly, exposure to examples did not systematically lead to fixation. Indeed, in the domain of
brainstorming, exposure to ideas from others can stimulate associations that lead to the generation of
additional ideas. In one experiment, Dugosh and Paulus [10] reported that participants generated more
unique solutions when a list of eight ideas was given prior to the brainstorming session. The neural
basis of this stimulation effect has been recently investigated in a functional Magnetic Resonance
Imaging (fMRI) study. Using a creative idea generation task where participants were requested to
generate alternative uses of conventional everyday object such as an umbrella, Fink et al. [3] reported
that creative performance increases after the exposure to examples of solutions. At a
neurophysiological level, cognitive stimulation with ideas from others is associated with activation
increase in a neural network involved in memory retrieval and attentional processes. Altogether these
results indicate an interaction between an external idea and participants’ memory and therefore
suggest that exposure to examples of a solution could modulate the bottom-up attention and retrieval
of novel associations, allowing participants to produce more original ideas. Nevertheless, these studies
provide no clues to aid us in understanding why examples lead to fixation effects in some situations
whereas in others they stimulate creativity.
2.2 Characterizing restrictive and expansive examples
According to us, theoretical work in design science provides insights on describing different features
of solutions to a design task. C-K design theory or concept-knowledge theory models the creative
process as the interrelated expansion of two spaces [4],[5]. One space (the Concept space) is tree
ICED 11
2
INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED11
15 - 18 AUGUST 2011, TECHNICAL UNIVERSITY OF DENMARK
structured and describes the progressive and stepwise generation of “temptative alternatives”, which
are usually indecidable propositions before a “conjunction” can be interpreted as “a solution”; the
other space (the Knowledge space) is formed by the network of memorized and activated knowledge
that is used for the generative process of the C-Space. One of the recognized breakthroughs provided
by C-K theory is that it reveals that to obtain a creative solution both processes have to reinforce each
other: thinking of an alternative changes the activated sets of knowledge and also vice versa.
C-K theory sets the framework of a design process based on refining and expanding the initial concept
by adding attributes stemming from the knowledge space. The initial concept-set is thus partitioned
step-by-step in several increasingly more refined sub-concepts. There are however two kinds of
partitions [4]. The restrictive partition is a partition where we restrict the space of possibilities without
changing the definition or the attributes of the object or process to design. However, sticking to
restrictive partitions does not allow the redefinition of objects; an expansive partition is a partition that
modifies the identity of the initial design object by adding unexpected attributes to that of the initial
concept. It is precisely because of those expansions that the true innovation, including surprises, is
possible.
In this context, C-K theory explains the fixation effect as arising from both a limitation of C and K
expansions. Interestingly, the theory predicts that people tend to generate ideas that are most
accessible in memory. This activation of common knowledge could lead to fixation effects.
Consequently, we hypothesized that external cues reinforcing the activation of these common
knowledge should increase the fixation effect. On the contrary, example of solutions that activate
knowledge that are less spontaneously accessible should reduce the fixation effect and therefore
stimulate creativity.
3 GENERAL PROCEDURE
Our investigation is divided in two phases: (1) identifying restrictive and expansive examples using CK reasoning, (2) examining the potential impacts on originality of both types of examples based on the
results of the first phase.
Therefore, we first used C-K theory to explore possible paths of innovation, capturing the possible
generation of new knowledge and objects on the matter of dropping an egg without breaking it. In a
first study we aimed to determine both fixation path based on spontaneous knowledge and expansive
path activating knowledge that is less accessible. In a pre-test session, a group of participants was
asked to give solutions to the egg task. We then used this natural distribution of solutions in order to
control that expansive solutions require an expansion of knowledge (ie the activation of a set of
knowledge that is not spontaneously activated when designing a solution to the egg task).
Then, using results of this first study, we identified restrictive and expansive examples. We then
examined their specific impact on originality in a second study, where three groups of participants
were asked to solve the egg task while given either of those examples. A fourth group was then given
both restrictive and expansive examples.
4 STUDY 1: METHOD TO IDENTIFY RESTRICTIVE AND EXPANSIVE
EXAMPLES
3.1 The use of C-K theory
To use C-K theory in our case, we first gathered the knowledge and expertise and the solutions usually
proposed on the topic (such as shells, mattresses, parachutes). Our hypothesis was that this knowledge
basis is the one that is classically activated while solving this task (see table 1). We then expanded in
the C-space by making explicit the hidden partition that had been chosen to achieve the design of
devices that change either the reception or the fall: those solutions design the drop of an egg from a
10-meter height without breaking it, using a device that is inert (eg not living). The expansions (a)
without using a device and (b) using a living device pushed us to expand in the K-space, to look for
new knowledge (on the egg properties, on living devices, etc), that then allowed us to think of new
expansions in the C-space (using the natural properties of the egg or modifying them).
Table 1: Sequences to build the C-K diagram
ICED 11
3
INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED11
15 - 18 AUGUST 2011, TECHNICAL UNIVERSITY OF DENMARK
Phase
Phase 1
Gathering of
spontaneous
knowledge
In the K-space
Model of dropping an egg from a 10
meters height.
Knowledge on gravity, fall, reception,
forces.
Benchmark of existing solutions:
shells, mattress, parachutes.
Some knowledge about the fragility of
the egg
Partition on device / with no device,
and inert device / living device.
Phase 2
Expansion in Cspace
Phase 3
Expansion in Kspace
Phase 4
Expansion in Cspace
And going
In the C-space
Choice between a focus on the
reception (damping the shock or
protecting the egg) or a focus on the
fall (slowing it)
Knowledge on the properties of the
egg, properties of the drop, of the
environment
Knowledge on living devices that
break falls, or that reception well.
Using the natural properties of the
egg, modifying them
…
…
The use of C-K theory allowed us to map the possible solutions axis, as it is shown on the Figure 1.
Thus, our hypothesis is that the use of C-K theory helped us to think of possible paths of solution that
do not come spontaneously to mind (training an eagle to catch the egg, using the natural robustness of
the longitudinal axis of an egg, freezing the egg, etc). We therefore examined in this first study
whether the paths of solutions generated by the C-K reasoning are more than the fixation effects paths.
3.2 Participants
Twenty-eight undergraduate students in psychology from Paris Descartes University were recruited
for this study. Subjects were between 17 and 22 years old (mean age: 18,8 years). Participants didn’t
have previous experience with design project.
3.3 Procedure
Each participant was given 10 minutes to generate as many original solutions to the following
problem:
“You are a designer and you are asked to propose as many original solutions as possible
to the following problem: Ensure that a hen's egg dropped from a height of 10m does
not break.”
3.4 Results
As illustrated in Figure 1, analysis of responses given by the participants revealed a strong fixation
effect. Indeed, 81 % of the responses consist in using an inert device in order to damp the shock (33
%), protect the egg (26 %), or slow the fall (22 %)(2 significant at p <.0005). One good example of
solutions in this fixation path based on common knowledge is to slow the fall with a parachute. On the
contrary, solutions belonging to expansive paths were less often proposed by the participants (2
significant at p <.0005). Interestingly, less than 10 % of the responses consist in using a living device
(4%) or in modifying the natural properties of the egg (3%). Examples of solutions belonging to these
expansive paths are to train an eagle to catch the egg during the fall or to freeze the egg before
dropping it.
ICED 11
4
INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED11
15 - 18 AUGUST 2011, TECHNICAL UNIVERSITY OF DENMARK
Figure 1 : C-K diagram for Study 1
ICED 11
5
INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED11
15 - 18 AUGUST 2011, TECHNICAL UNIVERSITY OF DENMARK
The key aim is now to determine whether examples belonging to the fixation path (restrictive
examples) and examples belonging to expansive paths (expansive example) have an opposite effect on
participants’ ability to generate creative ideas.
5 STUDY 2 : IMPACTS OF EXAMPLES ON ORIGINALITY
In the present study, we explored the impact of restrictive and expansive examples on creativity in the
egg task. Based on C-K theory, we hypothesized that the introduction of example during the egg task
would provoke participants to propose more original solutions when the provided example was
expansive (i.e. belonging to the expansive paths evidenced in the first study). In contrast, when the
example provided was restrictive (i.e. belonging to the fixation paths evidenced in the first study), we
hypothesized that the originality of the given solutions would be reduced. We have then conducted
two analyses, one on the mean number of solutions and one on an objective originality score of those
solutions.
4.1 Participants
One hundred and thirty two undergraduate students from Paris Descartes University were recruited for
this study. Subjects were between 17 and 28 years old (mean age: 19,1 years). Participants were tested
at the same time in a large room. All participants were naïve regarding the experimental aims and
didn’t have previous experience with design project.
4.2 Procedure
Each participant was randomly assigned to one of the four experimental conditions and was given 10
minutes to generate as many original solutions to one of the following problem.
 Control group (n=28): Without example
“You are a designer and you are asked to propose as many original solutions as possible
to the following problem: Ensure that a hen's egg dropped from a height of 10m does
not break.”
 Group A (n=25): Restrictive example
“You are a designer and you are asked to propose as many original solutions as possible
to the following problem: Ensure that a hen's egg dropped from a height of 10m does
not break. The most often proposed solution is to slow the fall with a parachute”
 Group B (n= 27): Expansive example 1
“You are a designer and you are asked to propose as many original solutions as possible
to the following problem: Ensure that a hen's egg dropped from a height of 10m does
not break. One possible solution is to train an eagle to catch the egg during the fall.”
 Group C (n= 26): Expansive example 2
“You are a designer and you are asked to propose as many original solutions as possible
to the following problem: Ensure that a hen's egg dropped from a height of 10m does
not break. One possible solution is to freeze the egg before dropping it.”
 Group D (n= 26): Restrictive example and Expansive example 1
“You are a designer and you are asked to propose as many original solutions as possible
to the following problem: Ensure that a hen's egg dropped from a height of 10m does
not break. Your company specializes in parachute slowing the fall of the eggs. A
competitor recently proposed as a solution the training of an eagle so that the eagle
catches the egg in the air.”
ICED 11
6
INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED11
15 - 18 AUGUST 2011, TECHNICAL UNIVERSITY OF DENMARK
4.3 Results
In order to examine whether the number of solutions proposed varied according to the experimental
conditions, we conducted one factor analysis of variance (ANOVA) with the experimental groups (A,
B, C, and D) as a between-subjects factor. This analysis revealed that main effect of experimental
groups is close to reach significance (F(4,129) = 2.09 p = .08), indicating that our experimental
conditions affect the number of solutions given by the participants. Interestingly, the two groups
exposed to restrictive examples proposed fewer solutions than the others groups (see Figure 2). It
should be noted that we measured the simple count of solutions generated and not a count of solution
types.
Figure 2 : Mean number of solutions according to the experimental groups
As usual in divergent thinking literature, we computed an objective measurement of originality of
solutions by taking the frequency of responses given across all the subjects in each group. In this
score, originality of a solution is defined as the statistical infrequency of that particular solution. For
example, if less than 5 % respondents report a response, that response is considered to be of high
originality. On the contrary if more than 95 % respondents report a response, that response is
considered to be of low originality. Thus, we obtained a mean originality score between 0 and 1 for
each participant where 0 represents the lower level of originality and 1 the higher one (see Figure 3).
These scores were subjected to one factor ANOVA with the experimental groups (A, B, C, and D) as a
between-subjects factor. This analysis showed a main effect of experimental groups (F(4,129) =
11.219, p < .0005), indicating that the presence of examples have a significant impact on the
originality of solutions.
ICED 11
7
INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED11
15 - 18 AUGUST 2011, TECHNICAL UNIVERSITY OF DENMARK
Figure 3 : Originality score according to the experimental groups
Post-hoc comparisons revealed that solutions proposed by group A are less original than those given
by the control group (p < .005), or group B (p <.005) and the group C (p <.005). This suggested that
the exposure to restrictive example reduces the originality of the solutions. On the contrary, our data
showed that the two groups exposed to expansive examples proposed more original solutions than the
control group without example (p = .07). Interestingly, participants who received both a restrictive and
an expansive examples are more original than those in group A (p <.0005) with restrictive example
solely. Note that there was no significant difference between group D and the two other groups with
expansive examples (all p’s >.05).
Beside our data also revealed that globally the first solution given by the participants is less original
than the last one (F(1,129) = 28.59, p < .0005) even if there is no significant interaction between
experimental groups and the first/last factor (F < 1).
6 DISCUSSION AND CONCLUSION
In the two studies, we explored the impact of restrictive (i.e. belonging to the fixation path) and
expansive (i.e. belonging to the expansive paths) examples on a creative task that is to design a way to
drop a hen's egg from a height of 10 m so that it does not break. Four major findings emerged from
this investigation: (i) participants’ ability to generate a creative idea is blocked by solutions that come
spontaneously to mind, leading to fixation effect (ii) the introduction of expansive example during the
egg task led participants to propose more original solutions (iii) in contrast, when the example
provided was restrictive, both the number and the originality of solutions were reduced; (iv)
interestingly when both a restrictive and an expansive examples were presented in the design brief,
participants gave fewer solutions without affecting the originality.
As predicted by C-K theory, our first study revealed strong fixation effects indicating that participants
gave solutions based on common knowledge spontaneously activated. This result is in keeping with
studies in the domain of reasoning that demonstrated the involvement of an intuitive/automatic system
in problem solving [12]. In other words, when people have to solve a problem in a creative way they
are constrained by automatic activation of common knowledge leading them to give usual solutions.
ICED 11
8
INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED11
15 - 18 AUGUST 2011, TECHNICAL UNIVERSITY OF DENMARK
Interestingly enough, when the activation of this common knowledge is reinforced by the presentation
of possible solutions belonging to the fixation path, participants proposed fewer solutions that were
also less original. This finding is in line with several previous works and confirms that recently
activated knowledge can strongly constrain the ability to generate creative solution [1], [8].
One major result of the present investigation is that expansive example produces the opposite effect on
originality. Indeed, providing example belonging to an expansive path in the C-space increases
creative performance. Although this result confirms that a stimulation effect of example is possible
[10], [3], see also [13], [14] it can appear contradictory with regards to the vast number of studies
mentioned previously reporting that examples block creativity. However thanks to C-K theory we can
suggest explanation of this apparent discrepancy. The exposure to an expansive example unveils a
possible expansion in the C-space to the participants, which is not the case for a restrictive example.
Thus, expansive example provides with a set of alternatives in the C-space: using a living device,
modifying the properties of the egg. This expansion could be however limited to the C-space in the
sense that participants did not have access to ways to expand their knowledge (for instance via the
internet). In addition our results suggest that participants did not transfer the fixation from the
common path to the expansive examples given in the present experiment.
A limitation of our study is that we computed an objective measurement of originality of
solutions without taking into account the appropriateness of the solutions given by the participants.
Yet, a creative idea is original, but an original idea is not always creative. Therefore, it would be
interesting in future works to use an external rating procedure similar to the Consensual Assessment
Technique (CAT) proposed by Amabile. Although the marked interest for the study of the impact of
external cues on creativity, there are many loose ends in the field and little is known about the neurocognitive processes allowing (1) to control routine strategies and (2) to generate new creative ones.
Using both new modeling of creative reasoning and recent advance in neuroimaging techniques it
would be possible in future work to directly observe the neural activity associated with (i) the fixation
effect (ii) the neuro-cognitive impact of both restrictive and expansive examples.
ICED 11
9
INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN, ICED11
15 - 18 AUGUST 2011, TECHNICAL UNIVERSITY OF DENMARK
BIBLIOGRAPHY
[1] Abraham A. and Windmann S. Creative cognition: The diverse operations and the prospect of
applying a cognitive neuroscience perspective, Methods, 2007, 42, pp38-48.
[2] Defeyter M.A. and German T.P. Acquiring an understanding of design: Evidence from children’s
insight problem solving. Cognition, 2003, 89, pp133–155.
[3] Fink A., Grabner R.G., Gebauer D., Reishofer G., Koschutnig K. and Ebner F. Enhancing
creativity by means of cognitive stimulation: Evidence from an fMRI study. NeuroImage, 2010, 52,
pp1687-1695.
[4] Hatchuel A. and Weil B. (2003). A New Approach to lnnovative Design : An Introduction to C-K
Theory. In International Conference on Engineering Design, ICED'03, Stockholm, August 2003, pp
14.
[5] Hatchuel A. and Weil B (2009). "C-K design theory: an advanced formulation." Research in
Engineering Design, 2009, 19(4), pp181-192.
[6] Duncker K. On problem-solving. Psychological Monographs, 1945, 58(5, Whole No. 270).
[7] Adamson R.E. Functional fixedness as related to problem solving: A repetition of three
experiments. Journal of Experimental Psychology, 1952, 44, pp288-291.
[8] Smith S.M., Ward T.B. and Schumacher J.S. Constraining effects of examples in a creative
generation task, Memory & Cognition, 1993, 21, 837-845.
[9] Jansson D.G. and Smith S.M. Design fixation, Design Studies, 1991, 12 (1), pp3-11.
[10] Dugosh K.L. and Paulus P.B. Cognitive and social comparison processes in brainstorming.
Journal of Experimental Social Psychology, 2005, 41, pp313-320.
[11] Cassotti M. and Moutier S. How to explain receptivity to conjunction-fallacy inhibition training:
Evidence from the Iowa Gambling Task. Brain & Cognition, 2010, 73 (3), pp378-384.
[12] Kahneman D. and Frederick S. Frames and brains: Elicitation and control of response tendencies.
Trends in Cognitive Sciences, 2007, 11(2), pp45-46.
[13] Ward T. B, Patterson M.J, & C Sifonis C.M. The role of specificity and abstraction in creative
idea generation. Creativity Research Journal, 2004, 16(1): 1-9.
[14] Howard T. J, Culley S.J, & Dekoninck A. The use of creative stimuli at early stages of industrial
product innovation. Research in Engineering Design, 2010, 21(4): 263-274.
ICED 11
10